Al's TAFE Certificate IV I.T. (Website Design) Exercises

Semester TWO

Lesson 13 » Produce and Manipulate Digital Images

Image resolution: sourcing files from scanners and digital cameras

Overview

This tutorial will take a look at how you can get the best possible image from a scan, and gives an understanding of the concepts of resolution and image resizing, particularly when sourcing files from scanners and digital cameras.

The traditional analog world was easy to understand. Negatives and prints were continuous tone and one simply made an enlargement from the negative or transparency to the size needed. Sharpness and grain were more of an issue rather than resolution. Today, in the digital environment, many newcomers get-tripped up by the concepts of input and output resolution and terribly confused as to what settings to use, and when.

To understand these issues clearly one needs to start with an acceptance of certain physical limitations of the human eye. Our vision is incapable of discriminating detail below a certain level. This varies from individual to individual and even by the same individual on different days (or perhaps it's dependent on how much sleep we've had!), but more or less this point is at about 200 dots per inch (dpi).

When an image is composed of dots smaller than this they appear to the eye as continuous tone. This has been relied upon by the printing trade for a couple of hundred years. Every photograph and every image that you see in every book, magazine, calendar and art reproduction is comprised of dots of ink, at resolutions typically ranging from 70 to 300 dots per inch.What you see in newsprint (or anything photomechanically reproduced) is affected by the number of pixels and the line screen/ dots used in the photomechanical process.

Digital images, whether derived directly from a digital camera or from a film scanner, obey the same laws. If the resolution used to make a print is too low we will "see the dots", just as you sometimes do in a photograph reproduced on low quality newsprint. 

In a digital photographic print, what you end up seeing (if the resolution isn't high enough) are the pixels. These are the discrete elements used to capture the image created by the camera or scanner lens system on the device's imaging chip. They are in effect the equivalent of the grain found in silver-based films or the dye clouds found in colour negative and transparency films. The problem comes when trying to understand the relationship between what is captured and what ends up on a print.

Optical Resolution

A scanner’s optical resolution is determined by how many pixels it can actually sample by its optics, mechanics and electronic circuits. For example, a typical flatbed scanning head with 300 sensing positions per inch, can sample 300 dots per inch (dpi) in one direction. To scan in the other direction, it moves the scanning head along the image, stopping 300 times per inch, so it can scan 300 dpi in the other direction as well. This scanner has, therefore, an optical resolution of 300 x 300 dpi. Some manufacturers stop the scanning head more frequently down the image, so their optical resolutions are, for example, 300 x 600 dpi or 300 x 1200 dpi. Don’t let these numbers mislead you; what is really important is the smallest number, you cannot get more detail by scanning more frequently in one direction only.

Interpolated Resolution

You should look out for claims about interpolated (or enhanced) resolution. Unlike optical resolution, which is a measure of the number of pixels that a scanner can actually sense, interpolated resolution is a measure of the number of pixels the scanner can guess at. Through a mathematical process called interpolation, the scanner turns, for example, a 300 x 300 dpi into a 600 x 600 dpi scan by inserting new pixels in between the optically sensed ones, and calculating at what light value it would have sampled in that position had it been there. In other words, the scanner inserts an in-between value based on the colours of the two adjacent pixels, the results are seldom satisfactory, and should therefore be avoided.

So interpolation is a fancy word for a computer "guessing". The only type of interpolation that anyone should be worried about is when an image is enlarged. Basically, when an image is enlarged, you end up with more pixels. But where did these pixels come from? Back in the old days, if you enlarged a 320x240 image to 640x480, the program would just double-up each pixel. Thus, the image looked blocky. Nowadays, with more advanced programs like Photoshop, interpolation can be done on the photo while enlarging. The interpolation calculations allow the computer to pick better pixels to add to the image rather than just using the pixel next to it. This way, when you enlarge an image, the image looks a lot smoother and cleaner. Note that the amount of photo information hasn't changed. If you look at an enlarged interpolated picture, you will see that if a person was a white dot because they were too far away, then they will still be a white dot only bigger. If you had an optical zoom then they would have looked more of like a person. But if you don't optical zoom, and then instead interpolate to get a closer view, it isn't going to change much.

Photoshop will interpolate upwards far better than a digital camera will. There are several plug-ins for Photoshop that are even better. The one that I've tried that works well is Extensis PXL Smart Scale. We'll talk about it in class.

What matters in digital cameras is the optical resolution too (other than lens performance like sharpness). "Digital zoom" is next to useless by the way - it simply reduces the resolution as it trims the image (and usually interpolates the image resolution upwards again to compensate).

Film Scanners

Film scanners often use very high scanning resolution at the literal size of the transparency or negative. Thus the image file then needs to be converted in Photoshop to a useable resolution at a larger output size. In class we will discuss - particularly with reference to the "resample image" check box in the "Image Size" dialog box:

What dpi should I use for digital photographic prints?

The minimum required for a decent print is 200dpi, but it's best to supply prints at  the preferred resolution of the destination printer. Richmond Instant Photo has a digital printer that has a printing resolution of 300dpi, and can do prints up to 16" by 12". F-Stop Online in Brisbane http://www.fstoponline.com.au/ has a Lambda printer capable of 50" (1.2m) high by any width and costs $0.006 per sq cm. They prefer 250dpi.

What resolution should I use for inkjet printers?

Use a multiple (or whatever the word is for the division multiple) of the printer's resolution. If it's a 720dpi printer (Epsons usually are related - like 1440dpi) then use 360 dpi or 180dpi. You would be surprised how good a print that is 90dpi can look on an Epson printer set at 360dpi (a multiple of 90).

What resolution for the web?

72dpi as you well know.

What resolution do I use for commercial ink press printing?

300dpi covers most uses. Newspapers and poorer quality mags are happy with 200dpi or even less - like 160dpi. There's a formula actually, based on the number of lines per inch required by the press:

dpi required = 1.5 x lpi

Exercise 1

Original pic resized and saved at high res 261KB


Saved as low resolution to 14KB


Re-saved from the low res to high res 70KB



Click here for the original pic in full size.

Exercise 2